Académique Documents
Professionnel Documents
Culture Documents
,
where AUM
i,t
and r
i,t
are, respectively, the assets under management and the return of
fund i at time t. Using these individual fund flows, we then compute the redemption
rate within investment style k and across all fund styles K k at time t as
8
Hedge funds rely on large investors money such as pension funds or high net-worth individuals to finance their
activity. The share of large institutional investors in the hedge fund capital has been rising over the last years and
reached more than 30% at the end of 2005 and is expected to increase further with the ageing of the population
(Casey et al., 2006).
16
ECB
Working Paper Series No 1112
November 2009
t k
t k
t k
USD TotalFlows
D NegFlowsUS
Within Redemption
,
,
,
=
t k K
t k K
t k K
USD TotalFlows
D NegFlowsUS
Across Redemption
,
,
,
=
=
=
=
where NegFlowsUSD
k,t
is the absolute value of the sum of negative flows at time t
considering all funds which belong to the style category k, while TotalFlowsUSD
k,t
denotes the sum of positive and the absolute value of negative flows of funds within
investment style k.
Failure channel: A cluster of hedge fund failures is also likely to have an impact on
the failure probability of existing funds. As hedge funds mostly rely on short-term fi-
nancing from prime brokers to pursue leveraged investment strategies (Greenwich
Associates, 2007), prime brokers are likely to tighten their credit conditions if the
failure rates of hedge funds suddenly increase. Since prime brokers have only incom-
plete information about the financial health of hedge funds, they may rationally take
into account an increase in the failure rate of hedge funds when forming their priors
about the risk of fund failure. As a consequence, prime brokers could then decide to
increase hedge funds margin requirements, reduce their maximum leverage limit or
credit lines, thereby propagating initial stress within the industry (Chowdry and
Nanda, 1998). As a result, hedge funds operating near their maximum leverage limit
could then be forced to simultaneously sell their assets in potentially unfavourable
market conditions (Ewerhart and Valla, 2007). According to Brunnermeier and Peder-
sen (2009), a margin spiral arises if higher margins increase the funding problems
of a hedge fund and therefore cause even higher margins. In addition, a loss spiral
arises as losses on a funds initial position force the fund to sell more of its assets
which causes a further decline in prices. Both spirals reinforce each other leading to a
combined effect which is larger than the sum of their individual effects. Declining
fund returns and a rising failure probability could then follow. The effect should be
more pronounced if failures occur within funds having similar risk and market expo-
sures than across funds from different investment styles (Cifuentes et al. 2005).
17
ECB
Working Paper Series No 1112
November 2009
Hypothesis 3: All else equal, in the presence of incomplete information, an increase in
the failure rate of hedge funds serves as a signal for the deterioration of the financial
health of hedge funds. As a consequence, prime brokers should tighten the credit con-
ditions for existing hedge funds which is related to an increase in their failure prob-
ability due to a heightened funding risk.
We measure the failure rate of hedge funds within investment style k and across all
fund styles K k at time t as
t k
t k
t k
nds ExistingFu
s FailedFund
eWithin FailureRat
,
,
,
=
t k K
t k K
t k K
nds ExistingFu
s FailedFund
eAcross FailureRat
,
,
,
=
=
=
=
where FailedFunds
k,t
and ExistingFunds
k,t
are, respectively, the number of fund fail-
ures during month t and the number of existing funds at the end of month t in style
category k.
4. The role of portfolio diversification in risk spillover
A policy issue of interest is to investigate whether hedge funds with well diversified
investment portfolios tend to have lower failure probabilities than less diversified
funds in both tranquil and crisis periods. In tranquil periods, we could expect that
diversification increases the survival probability, as empirical evidence tends to docu-
ment positive international diversification benefits, when emerging markets (EM) are
involved.
9
Therefore, the variance of hedge fund portfolios could be reduced for a
given level of expected returns when extending the geographical focus to EM, even
after accounting for their short-sale constraints and high transaction costs (Li et al.,
9
The early studies that ignored short-sale constraints and market frictions have documented low correlation across
international markets and some diversification benefits (Harvey, 1995; Bekaert and Urias, 1996; De Santis and
Gerard, 1997). For those studies that account for the short-sale constraints, empirical evidence has been rather
mixed. On the one hand, De Roon et al. (2001) showed that after major liberalisation in emerging markets, diversi-
fication benefits vanish once such market frictions are taken into account. On the other hand, Li et al. (2003) show
that international diversification benefits remain substantial for US equity investors even when they are prohibited
from short-selling in emerging markets, while Driessen and Laeven (2007) also find that global diversification
benefits remain large after controlling for short-sales constraints in developing stock markets.
18
ECB
Working Paper Series No 1112
November 2009
2003). Although hedge funds used to prefer rather liquid markets as they may exit
them rapidly without incurring prohibitive costs of liquidation, a significant propor-
tion of funds invest in emerging markets (59%).
10
We could thus expect that there are
positive benefits from international diversification in tranquil periods. On the other
hand, during financial turmoil, diversified portfolios could also amplify the effects of
risk spillover, potentially offsetting the positive benefits of diversification (Butler and
Joaquin, 2002). The final impact is therefore undetermined.
To assess the effect of diversification in hedge fund failure patterns, it is crucial to
measure diversification correctly in the first place. The use of the TASS database for
this purpose raises several concerns that should be considered when interpreting our
results. Each fund reports in which geographical areas it invests, for example in India
and in the US. However, no quantitative data are available to gauge the proportion of
investment in each region.
11
The same caveat applies for the type of assets: hedge
funds only report whether they invest in equities, fixed-income, currencies, commodi-
ties or property (or a combination of these assets). However, as hedge funds provide
information on their investment focus to TASS to attract new investors, it seems rea-
sonable to argue that the information declared by the manager of the fund are accu-
rate. Based on the available information, the evaluation of the degree of diversifica-
tion of hedge fund portfolios can only be qualitative. Several sets of dummy variables
are introduced to classify hedge funds as either diversified from a geographical per-
spective or diversified in terms of asset types or diversified according to the two
dimensions or not diversified at all.
12
10
Based on the information given in the TASS database, 10% of hedge funds invest only in EMEs, 10% disclose
to invest both in EMEs and in mature markets, while 39% declare to invest globally, which has been treated as
involving an investment in EMEs.
11
In the TASS database, there are seven emerging market areas (Africa, Asia-Pacific excluding Japan, Asia-
Pacific, Eastern Europe, India, Latin America and Russia), while there are seven mature geographical focuses
(United States, Japan, North America, United Kingdom, Western Europe, Western Europe excluding UK, North
America excluding the US).
12
As we use dummy variables to define diversification of a hedge fund, the effect on fund failure might be over-
stated. This aspect should therefore be taken into account when interpreting our results. An alternative would be to
determine the exposure of each fund to different regions and assets classes (and thereby the degree of diversifica-
tion) using the beta coefficients of style-type regressions. However, this approach may also have some drawbacks,
in particular as hedge funds often have nonlinear exposures to one or several asset classes which could make it
difficult to identify their degree of diversification (see, e.g., Agarwal and Naik, 2004).
19
ECB
Working Paper Series No 1112
November 2009
Geographical diversification dimension: Our definition of geographical diversifica-
tion rests on the distinction between mature and emerging markets. One could expect
a low correlation between asset returns from EMEs and mature economies and there-
fore positive diversification benefits reducing the failure probability of hedge funds.
The dummy variable equals 1 if a hedge fund invests simultaneously in mature and
emerging markets and 0 otherwise.
13
About 49% of hedge funds in the filtered data-
base report that they invest in these two geographical areas, while about 41% only
invest in mature economies and 10% only in emerging market economies.
Assets diversification dimension: A portfolio is assumed to be diversified if it encom-
passes at least two different types of assets, such as equities and bonds. In a first step,
the asset diversification dummy is equal to 1 in this case and 0 otherwise. Further-
more, we want to measure the potential different benefits of diversification from an
initial situation where portfolios are either composed of equities only or of bonds
only. Therefore we define an additional variable that has three outcomes: (i.) invested
only in equities, (ii.) invested only in bonds, (iii.) diversified in equities and bonds.
14
In the filtered database, about 86% of hedge funds report that they invest in equities,
while 45% invest in both equities and bonds.
Both geographical and asset diversification: Another set of dummy variables is intro-
duced, which combines the two previous diversification dimensions. Hedge funds are
then classified as (i.) diversified in the two dimensions (36%), (ii.) not diversified at
all (27%), (iii.) diversified in assets (24%) or (iv.) geographically diversified (13%).
Hypothesis 4: All else equal, on the one hand, hedge funds should be able to lower
their portfolio risk through diversification across different geographical regions
and/or asset classes. This should be related to a lower failure probability compared
to funds not being diversified. On the other hand, risk spillovers are likely to be am-
plified through diversification which should be associated with a higher failure prob-
13
Hedge funds which report to follow a global investment strategy are classified as diversified in our definition,
while those funds that do not provide any information on this issue or that report to invest in other areas have
been discarded from the analysis.
14
We ignore two other possible categories: (i.) not diversified in the other asset classes like commodities or cur-
rencies and (ii.) diversified in other categories than equities and bonds. Since the non-diversified portfolios in these
asset classes account for a very small proportion of hedge funds, we dont report the estimation results for these
categories. Empirical results are available upon request.
20
ECB
Working Paper Series No 1112
November 2009
ability. Therefore, due to these two competing effects, the overall effect of diversifica-
tion on hedge funds failure probability is a priori unknown.
5. Empirical results
This section aims at assessing whether the spillover of risk from one hedge fund to
another had a significant impact on funds failure probability over the period January
1994 to May 2008.
15
It also intends to gauge the impact of portfolio diversification on
the spillover of risk: is risk spillover more likely in the presence of diversified invest-
ment portfolios as opposed to concentrated market positions? The modelling of the
failure process of hedge funds and estimation of the baseline model are described
first. Then, the empirical results of the two channels of risk spillover are presented.
Finally, the effect of portfolio diversification on funds mortality patterns is investi-
gated.
5.1. Baseline model
Our specification of the hedge fund failure process uses a binary logit model, which
explains the outcome of a continuous latent variable
*
,t i
y , representing the unobserved
failure probability of fund i at time t, by a matrix of explanatory variables
i
x :
16
t i i t i
y
,
*
,
c + = x .
(1)
To estimate this model we use a sample of N hedge funds i = {1, 2, , N}, observed
over T periods t = {1, 2, , T}. Each hedge fund i in every month t is classified as
either failed or alive.
17
This information is specified by the indicator variable y
i,t
,
which is linked to the latent failure probability
*
,t i
y in the following way:
15
TASS classifies funds that stopped reporting as Graveyard after a period of 8 to 10 months. To reduce the risk
of misclassification, the sample discards June to July 2008 data. Reducing the sample further does not significantly
change the results.
16
As some explanatory variables enter equation (1) with lags and to simplify notation, the time index t is omitted.
When we examine the robustness in section 6, we also report the results from a Cox proportional hazards model
which are similar to those of the logit model.
17
We follow Gregoriou (2002) and treat Graveyard funds not classified as Liquidated as censored at the date of
last report, rather than as failed at that date. To examine the robustness of our results with regard to this specifica-
tion, we repeat our analysis with a sample that (i.) discards all Graveyard funds not classified as Liquidated and
21
ECB
Working Paper Series No 1112
November 2009
s
>
=
) 0 if ( otherwise 0
0) (if month in liquidated is fund if 1
*
,
*
,
,
t i
t i
t i
y
y t i
y .
(2)
The probability that hedge fund i fails at time t, conditional on the values of the ex-
planatory variables, is then given by:
) exp( 1
1
) | 0 Pr( ) | 1 Pr(
*
, ,
x
x x
i
i t i i t i
y y
+
=
> = =
.
(3)
The baseline model is specified as follows:
t i
h
t i h h
h
t i h h i t i
style I year I y
,
10
1
, ,
14
1
, ,
*
,
) ( ) ( c | o + + + + =
= =
x ,
(4)
where the vector of hedge fund characteristics
i
x includes age, returns, size and capi-
tal flows as well as information on the use of leverage, the fees structure, cancellation
policy, minimum investment and investment in derivatives. In order to account for
economy-wide effects, indicator variables for 14 out of 15 calendar years denoted by
) (
, , t i h
year I and for 10 out of 11 investment styles denoted by ) (
, , t i h
style I are in-
cluded.
18
We also include a December dummy to capture the fact that fund manage-
ment companies tend to close their funds towards the end of the year. We expect a
negative sign for the coefficients of age, returns, size and capital flows since older,
more successful, bigger funds and those with larger inflows are less likely to fail
(Chan et al., 2005 and Baquero et al., 2005). Moreover, we expect a high watermark,
low redemption frequency, high redemption notice, payout and lockup periods to re-
duce the probability of fund failure (Goetzmann et al., 2003; Brown et al., 2001; Pan-
ageas and Westerfield, 2008).
(ii.) classifies funds as Liquidated if their aggregated capital flows over the last twelve months preceding disap-
pearance are negative. The results presented in section 6 are similar to those in this section.
18
We also included financial control variables such as the 3-month US Treasury bill rate, broad stock and bond
indices, but they turned out to be not significant. Therefore, they have been dropped from the list of regressors.
22
ECB
Working Paper Series No 1112
November 2009
The logit models are estimated by maximum likelihood through pooled regressions to
address the right censoring issue.
19
In order to ease the convergence during the esti-
mation and simplify the comparison of coefficients, all non-indicator variables have
been standardised to have a zero mean and a standard deviation of one. The results of
the baseline model are reported in the first column of Table 3. As expected, larger
fund returns, size and capital flows reduce the probability of fund failure.
20
This is in
line with the findings of Chan et al. (2005) and mostly consistent with Baba and Goko
(2006) and Baquero et al. (2005). Regarding the fee structure, management fees prove
to be insignificant, while higher incentive fees increase the failure probability of
hedge funds. Furthermore, we find that a high watermark lowers the failure probabil-
ity, which could reflect the incentive for fund managers to implement less risky in-
vestment strategies. Focusing on cancellation policy, our empirical results reveal that
the longer the lockup, redemption notice and payout periods, the lower the failure
probability. Furthermore, the industry practice to close a hedge fund at the end of the
year seems to be reflected in the large positive coefficient of the Month 12 dummy.
Finally, investment in options and swaps tends to be related to a higher failure prob-
ability.
5.2. Risk spillover within and across hedge fund styles
This section presents the procedure for testing the presence of risk spillover within
and across hedge fund categories as well as the empirical results regarding the sig-
nificance of the redemption and failure channel. These two channels are tested sepa-
rately in equation (5a) and then simultaneously in (5b):
19
A right censoring issue generally arises in survival analysis as most of the individuals observed over a given
period did not experience the event, which requires maximum likelihood estimation.
20
Following Chan et al. (2005) and Baba and Goko (2006) we include only assets under management in t-1 in the
estimation to avoid multicollinearity problems. Using assets under management in t instead yields almost identical
results.
23
ECB
Working Paper Series No 1112
November 2009
( )
( )
t i
k j
a k
j t i
a k
j
w k
j t i
w k
j
h
t i h h
h
t i h h i t i
t i
j
a k
j t i
a k
j
w k
j t i
w k
j
h
t i h h
h
t i h h i t i
c c
style I year y
k
c c
style I year y
,
2
1
2
2
,
,
, ,
,
,
10
1
, ,
14
1
, ,
*
,
,
2
1
,
,
, ,
,
,
10
1
, ,
14
1
, ,
*
,
) (
2 , 1 for
) (
c
| o
c
| o
+ + +
+ + + =
=
+ + +
+ + + =
= =
= =
=
= =
x
x
(5a)
(5b)
where
s k
j t i
c
,
,
defines the variable capturing channels of risk spillover, with k=1 (k=2)
referring to the redemption (failure) variable, and with s=w (s=a) denoting risk spill-
over within (across) hedge fund styles corresponding to fund i at time t-j. The coeffi-
cient
w k
j
,
(
a k
j
,
) measures the spillover effect within (across) hedge fund styles in
month t-j. As investor redemptions can only be observed by market participants once
they are reported in TASS (or in the press for well-known funds) it seems reasonable
to argue that the failure probability of hedge funds is affected only one or two months
after the initial redemptions within or across fund styles occurred. In the case of fail-
ures of other funds, it is also likely that it takes some time until prime brokers tighten
their credit conditions and the failure probability of existing funds is affected. There-
fore, we only include lagged values of our spillover variables in the models.
The estimation results on risk spillover are presented in columns 2 to 4 of Table 3.
Column 2 shows that the higher the redemption rate within hedge funds of one in-
vestment style in t-2, the higher the likelihood that a fund belonging to the same style
category fails in t. This result is economically significant. If the redemption rate
within a certain hedge fund style increases from its sample mean by one standard de-
viation, the monthly failure probability of a hedge fund increases by about 11% rela-
tive to its baseline failure probability. On the other hand, a higher redemption rate
across hedge funds from different styles in t-1 or t-2 is related to a lower failure prob-
ability in t of a fund not belonging to the style category where the initial redemptions
occurred. If the redemption rate across hedge fund styles in t-1 (t-2) increases by one
standard deviation, the monthly failure probability of a hedge fund decreases by about
13% (9%) relative to its baseline failure probability.
24
ECB
Working Paper Series No 1112
November 2009
Regarding the second channel of risk spillover, column 3 (Table 3) shows that a rise
in the failure rate of hedge funds within one investment style in t-1 or t-2 is associated
with an increase in the failure probability of an existing fund belonging to the same
style category in t. These results are economically significant. If the failure rate of
hedge funds within a particular style category in t-1 or t-2 increases by one standard
deviation, the failure probability of an existing fund in the same style increases by
about 5% relative to its baseline failure probability. There is no significant relation
between the failure rate across funds from different investment styles and the failure
probability of existing funds not belonging to one of these style categories. Compared
with the redemption channel, the effect of an increase in the failure rate on the failure
probability of existing funds is smaller. This seems plausible as the transmission
mechanism (fund failures being recognized by prime brokers which tighten hedge
funds credit conditions and therefore increase their funding risk) is supposed to be
weaker in the case of fund failures compared to that of investor redemptions.
When testing simultaneously the relevance of the two spillover channels, the coeffi-
cients are almost unchanged. Finally, we also test and reject the null hypothesis
{ } { } { } 2 , 1 , 2 , 1 , , , 0 :
,
0
= = = = k j a w s H
s k
j
that the coefficients on redemption
ratios and failure rates within and across hedge fund styles are simultaneously non-
significant.
21
Spillover effects from the two identified channels have therefore signifi-
cantly increased the failure probability of hedge funds over the period under consid-
eration. Although rejecting the null hypothesis is not a formal proof of risk spillover,
it is at least consistent with the presence of spillover effects within and across hedge
fund categories. Overall, these findings provide strong support to our hypotheses 1, 2
and 3.
5.3. Effect of diversification
This section aims at evaluating whether spillover effects are magnified or reduced if
investment portfolios of hedge funds are diversified. We might expect that well diver-
sified portfolios reduce the failure probability of hedge funds in quiet periods,
21
The value of the corresponding likelihood ratio (LR) test is given in the lower part of Table 4.
25
ECB
Working Paper Series No 1112
November 2009
whereas it could increase failure potential during periods of financial stress. We first
analyse the impact of diversification on failure probability, before introducing risk
spillover channels in the estimation.
5.3.1. Effect of diversification on failure probability
The impact of diversification on failure probability of hedge funds is investigated in
three stages. First, we evaluate the effect of the two diversification dimensions (geog-
raphy and assets) separately. That is,
2 , 1 for
) (
,
10
1
, ,
14
1
, ,
*
,
=
+ +
+ + + =
= =
d
div
style I year y
t i
d d
h
t i h h
h
t i h h i t i
i
c o
| o x
(6a)
where the dummy variable
d
i
div equals 1 if hedge fund i is diversified and 0 other-
wise. The superscript d refers to the diversification dimension, with d=1 for geogra-
phy and d=2 for the type of assets. As expected, a diversified portfolio tends to reduce
the failure probability of hedge funds, suggesting the presence of some diversification
benefits (Table 4). Such findings hold for a portfolio diversified from a geographical
or from an assets perspective, with the geographical diversification between mature
and emerging markets having the smallest impact in lowering the failure probability
(columns 1 to 2). These results are economically significant. If a hedge fund is diver-
sified from a geographical perspective (in terms of assets), its failure probability is
about 17% (19%) lower relative to its baseline failure probability than for a fund
which is not diversified at all.
Second, we investigate in more detail the diversification dimension related to the asset
types. We distinguish three possible states for a portfolio: (i.) invested in equities only
(
1
i
adiv ), (ii.) invested in bonds only (
2
i
adiv ), (iii.) diversified in equities and bonds
(
3
i
adiv ). Since the asset diversification dimension has three outcomes, each of these
dummies is used once as a reference category in the estimation, displayed in columns
3 to 5 in Table 4. As an example, the following equation uses
1
i
adiv as the reference
category:
26
ECB
Working Paper Series No 1112
November 2009
t i i i
h
t i h h
h
t i h h i t i
adiv adiv
style I year y
,
3 3 2 2
10
1
, ,
14
1
, ,
*
,
) (
c q q
| o
+ + +
+ + + =
= =
x
,
(6b)
where the coefficient
3
q measures the effect on the failure probability of a shift from
a portfolio composed of only equities to a diversified portfolio in assets. According to
the estimation results, a fund with a non-diversified portfolio would significantly re-
duce its failure probability by increasing the number of assets in its portfolio. Such
positive effect appears to be slightly larger when the fund has initially invested in
bonds (-0.326, column 4 compared to -0.254, column 3). These results are also eco-
nomically significant. If a hedge fund is initially invested only in equities (bonds) and
decides to invest both in bonds and equities, this is related to a decrease in the failure
probability of the fund by about 22% (28%) relative to its baseline failure probabil-
ity.
Third, the combination of the two diversification dimensions assets and geography
is analysed. A new variable is introduced, which has four outcomes: (i.) diversified in
assets and geographically (
1
i
cdiv ), (ii.) not diversified at all (
2
i
cdiv ), (iii.) diversified
in assets only (
3
i
cdiv ), (iv.) diversified geographically only (
4
i
cdiv ). As an example,
when
2
i
cdiv is taken as the reference category, the specification turns to:
t i i i i
h
t i h h
h
t i h h i t i
cdiv cdiv cdiv
style I year y
,
4 4 3 3 1 1
10
1
, ,
14
1
, ,
*
,
) (
c k k k
| o
+ + + +
+ + + =
= =
x
,
(6c)
where
p
k measures the effect of increasing the degree of diversification from a situa-
tion where the portfolio is initially not diversified at all. The effect of diversification
on failure probability when both the geographical and the asset dimensions are com-
bined is presented in columns 6 to 9 in Table 4. The largest diversification benefits
arise when a fund is not diversified at all and decides to diversify its portfolio both
geographically and in terms of assets (-0.345, column 7). Interestingly, as mentioned
before, the benefits stemming from geographical diversification (-0.125, column 7)
appear lower than those from diversification in assets (-0.169, column 7). These re-
27
ECB
Working Paper Series No 1112
November 2009
sults are economically significant. If a hedge fund is not diversified at all and decides
to diversify its portfolio both geographically and in terms of assets (only geographi-
cally; only in terms of assets), this is related to a decrease in the failure probability of
the fund by about 29% (12%; 16%) relative to its baseline failure probability. Test-
ing the null hypothesis { } 4 , 3 , 1 , 0 :
0
= = p H
p
k reveals that the diversification of
portfolios reduces the failure probability of hedge funds significantly.
22
Overall, these
results provide strong support to the first part of hypothesis 4 which states that diver-
sification across regions and or asset classes helps to reduce hedge funds failure
probability.
5.3.2. Effect of diversification on risk spillover
In a next step, interaction variables are introduced combining risk spillover with di-
versification to test whether spillover effects are amplified through diversification.
That is,
| |
( ) | |
t i
k j
i
a k
j t i
a k
j
w k
j t i
w k
j
k j
a k
j t i
a k
j
w k
j t i
w k
j i
h
t i h h
h
t i h h i t i
div c c
c c div
style I year y
,
2
1
2
1
,
,
, ,
,
,
2
1
2
1
,
,
, ,
,
,
10
1
, ,
14
1
, ,
*
,
) (
c
| o
+ + +
+ + +
+ + + =
= =
= =
= =
x
(7)
where, to simplify matters, we dont differentiate between the types of diversification
but instead use the dummy variable
i
div which equals 1 if hedge fund i is diversified
both geographically and in terms of assets and 0 otherwise. A positive
w k
j
,
or
a k
j
,
refers to spillover effects being amplified by the presence of a diversified portfolio.
The overall effect of diversification on failure probability is a priori undetermined.
On the one hand, we expect diversification itself to reduce the failure probability, i.e.
0 < . On the other hand, being diversified could increase risk exposures and there-
fore the probability of being affected by risk spillover.
22
The corresponding value of the LR-test is given in the lower part of Table 4.
28
ECB
Working Paper Series No 1112
November 2009
The results of the model with interaction variables are reported in Table 5. They re-
veal that, all else being equal, a hedge fund which is diversified both in terms of assets
and geographically has a significantly lower failure probability than a fund not diver-
sified in both types (-0.215). This provides again support to the first part of hypothesis
4. As we have included interaction terms, the results of the variables capturing spill-
over effects refer to hedge funds that are not diversified in both types. They are very
similar to the results reported in column 4 of Table 3.
Regarding the results of the interaction variables, two coefficients, which are related
to investor redemptions, are significantly different from zero indicating that the de-
gree of diversification only affects the impact of risk spillover on hedge funds failure
probability via redemptions.
First, the coefficient of interacting redemption within in t-2 with the diversification
dummy is negative (-0.153) and slightly significant. This implies that a hedge fund
which is diversified both in assets and geographically suffers to a smaller extent from
risk spillover (via redemptions within fund styles) than a fund that is not diversified in
both types, which provides support to the first part of hypothesis 4. Overall, the im-
pact of redemptions occurring within one investment style in t-2 on the failure prob-
ability of a (diversified) fund in the same style category in t is close to zero (0.152-
0.153). A possible explanation for this finding might be that diversified funds are per-
ceived as being different from the other funds of one specific investment style and
thus do not suffer so much from redemptions occurring within funds of one style cate-
gory.
Second, the coefficient of interacting redemption across in t-1 with the diversification
dummy is positive (0.245) and significant. This implies that if a hedge fund is diversi-
fied in both types its failure probability is almost not affected (-0.218+0.245) through
redemptions occurring across funds from different investment styles. In section 4 we
argued that in the presence of redemptions across funds from different styles those
funds belonging to the style category not affected by the capital outflows might bene-
fit from investors capital reallocations. This result, however, indicates that diversified
funds might be perceived as being more similar to funds from the other investment
styles which prevent investors from shifting their capital to them.
29
ECB
Working Paper Series No 1112
November 2009
Overall, these results imply that a hedge fund which is diversified in assets and geo-
graphically is not significantly affected by risk spillover through redemptions occur-
ring within and across fund styles. This means that, in contrast to the second part of
hypothesis 4, hedge funds seem to benefit from diversification (in addition to the ef-
fect captured by the diversification dummy) in the sense that diversified funds are not
affected by risk spillover via investor redemptions.
Finally, we test the null hypothesis that the impact of risk spillover on failure prob-
ability does not depend on the level of diversification. That is, we test whether the in-
teraction terms provide no significant additional explanatory power, compared to the
model including the baseline scenario and the variables capturing spillover effects and
diversification. The corresponding likelihood-ratio test displayed in the lower part of
Table 5, leads to a rejection of the null hypothesis at the 5% level.
6. Robustness
This section intends to examine the robustness of our results with regard to three as-
pects. The first aspect relates to the specification of hedge fund failure used in our
analysis. Following Gregoriou (2002), we treat Graveyard funds that stopped report-
ing to TASS but are not classified as Liquidated as censored at the date of last re-
port, rather than as failed at that date. However, a subset of those funds might still
have stopped reporting to TASS due to failure even if TASS did not classify them as
Liquidated due to a lack of information. To differentiate between funds that stopped
reporting because of failure or due to self-selection, we follow the approach of
Baquero et al. (2005) and classify funds as Liquidated if the aggregated capital
flows over the 12 months preceding the disappearance are negative. Based on that,
about 56% of the Graveyard funds not classified as Liquidated have negative ag-
gregated capital flows and are therefore classified as Liquidated while the remaining
funds are treated as censored. The corresponding estimation results are reported in
row 1 of Table 6 and are qualitatively similar to those in Table 3. The fact that, com-
pared to the results of Table 3, some of the coefficients are significant at lag two in-
stead of lag one or vice versa does not change our key result that investor redemptions
and a large number of failures within funds of one investment style increase the likeli-
hood of failure of an Alive fund belonging to the same style category.
30
ECB
Working Paper Series No 1112
November 2009
As an alternative to the approach of Baquero et al. (2005), we discard those Grave-
yard funds not classified as Liquidated and repeat the estimation with a sample
which therefore only consists of Alive and Liquidated funds. The results shown in
Table 6, row 2 are very similar to those reported in Table 3. Both findings show that
our key results are robust with respect to the specification of Alive and Liquidated
funds.
The second aspect in the context of checking the robustness of our results relates to
the model that we use for analysing hedge fund failure. While the binary logit model
is widely used in the context of survival analysis (see, e.g., Campbell et al., 2009;
Chan et al., 2005; Getmansky, 2005), several authors argue that the semiparametric
Cox proportional hazards model has some advantages such as not imposing a particu-
lar functional form for the dependence of a funds hazard rate on its age (see, e.g.,
Lunde et al., 1999; Brown et al., 2001). To account for this aspect, we repeat the
analysis using a semiparametric Cox hazards rate model. The estimation results are
reported in row 3 and 4 of Table 6.
23
Using the Cox model yields very similar results
as the logit model. It confirms that our results are robust to the way we have modelled
hedge fund failure.
Finally, we check whether our results are robust to using risk-adjusted rather than raw
returns as a control variable capturing hedge fund performance. We use Sharpe ratios
computed with a 6-month rolling window as our measure of risk-adjusted returns. The
estimation results are reported in row 5 of Table 6 and are very similar to those re-
ported in Table 3. This confirms that our results are robust to different specifications
of hedge fund performance used as a control variable.
7. Policy implications
This section aims at providing a list of variables that have a significant impact on
hedge fund failure and for which disclosure might be valuable from a transparency
and financial stability perspective.
23
The results of the other models are available upon request.
31
ECB
Working Paper Series No 1112
November 2009
To isolate the impact of individual explanatory variables on the failure probability of
hedge funds, we compute the marginal effect of a one standard deviation shock to
each regressor, all other variables being unchanged. An initial failure probability is
calculated first, where all explanatory variables are set to their sample mean and the
dummy variables to zero. As our models include yearly fixed effects, we have to spec-
ify a year for which the failure probabilities are computed. We thus assume that the
exemplary hedge fund operates in 2007. This yields a monthly baseline failure
probability of 0.20% (column 1, Table 7). The failure probability resulting from each
explanatory variable being shocked by one standard deviation is reported in column 2.
The marginal effect of each regressor is calculated as the difference between the two
failure probabilities (column 3).
24
In addition, we provide the 95% confidence interval
of the marginal effect in columns 4 and 5. As a result, hedge fund size, capital flows,
high watermark provision and the funds redemption notice period have the largest
individual effect in reducing the failure probability. Other hedge fund characteristics
also dampen the risk of hedge fund failure, albeit with a lower impact: contemporane-
ous and past returns, payout and lockup periods as well as redemptions across hedge
fund styles. On the other hand, the variables contributing to a significant increase in
failure probability are investment in derivatives, redemptions within funds from one
style category, incentive fees and failures among funds within one investment style.
In the recent debate on hedge fund regulation several authors have argued that at least
systemically important hedge funds should provide more transparency to regulators
on a confidential basis (see, e.g., Lo, 2008). Based on our results, it seems to be im-
portant that hedge funds provide information on size, capital flows, restriction peri-
ods, incentive fees and on their investment into derivatives to regulatory authorities.
In particular, information on capital flows and on a funds restriction periods is impor-
tant for enabling a regulator to evaluate the funding risk of a hedge fund and thereby
its failure probability.
24
For dummy variables the marginal effects are calculated by setting their values to one in the stress case.
32
ECB
Working Paper Series No 1112
November 2009
8. Conclusion
This paper analyses the determinants of hedge fund failures over the period January
1994 to May 2008. In addition to fund specific factors, we investigate whether a spill-
over of risk from one fund to another affects the failure probability of hedge funds. In
particular, we test two different channels through which a risk spillover might materi-
alise: investor redemptions and failures of other funds. Furthermore, we differentiate
between a spillover of risk taking place within funds from one specific investment
style and across funds from different style categories.
We find that, in addition to hedge fund characteristics, a spillover of risk from one
fund to another has a significant impact on the failure probability of hedge funds.
Comparing the two channels of risk spillover, we find that investor redemptions have
a larger impact on funds failure probability than failures of other hedge funds.
Our results also show that funds within the same investment style are adversely af-
fected through both channels of risk spillover: a rise in both the redemption and the
failure rate significantly increases the probability of fund failure. The result suggests
that changes in both variables are recognized by market participants such as investors
and prime brokers and are considered in their decision-making process, for example
when forming their priors on the riskiness of hedge funds.
On the other hand, hedge funds across different style categories are only affected
through the redemption spillover channel: an increase in the redemption rate reduces
the failure probability of funds operating in the style category not affected by the re-
demptions, i.e. it is actually beneficial for the corresponding funds. This finding indi-
cates that investors reallocate their capital to funds which are perceived as superior
since they did not suffer from outflows in previous months.
This paper also analyses whether and to which extent portfolio diversification affects
the failure probability of hedge funds. As the TASS database provides only qualitative
information on diversification of hedge funds investment portfolios, we use different
definitions of diversification. Overall, we find that hedge funds being diversified ei-
ther in terms of assets or geographically have a significantly lower failure probability
than funds being invested in just one asset class or one geographical region. In addi-
33
ECB
Working Paper Series No 1112
November 2009
tion, we investigate whether diversification amplifies the impact of risk spillover
among hedge funds. We find that the degree of diversification only affects the impact
of risk spillover on hedge funds failure probability via investor redemptions. In par-
ticular, hedge funds seem to benefit from diversification (in addition to the effect cap-
tured by the diversification dummy) in the sense that diversified funds are not affected
by risk spillover via investor redemptions.
Finally, the variables that have the largest impact on hedge funds failure probability
and which should therefore be disclosed on a confidential basis to regulators encom-
pass information on funds size, capital flows, restriction periods, incentive fees and
on their investment into derivatives.
34
ECB
Working Paper Series No 1112
November 2009
References
Adrian, T., Brunnermeier, M.K., 2009. CoVar. Working Paper.
Agarwal, V., Naik, N.Y., 2004. Risks and portfolio decisions involving hedge funds.
Review of Financial Studies 17 (1), 63-98.
Allen, F., Gale, D., 2000. Financial contagion. Journal of Political Economy 108 (1),
1-33.
Baba, N., Goko, H., 2006. Survival analysis of hedge funds. Working Paper Series
No. 06-E-05, Bank of Japan.
Bae, K.H., Karolyi, G.A., Stulz, R.M., 2003. A new approach to measuring financial
contagion. Review of Financial Studies 16 (3), 717-763.
Banque de France, 2007. Financial stability review: Special issue on hedge funds, No.
10, April.
Baquero, G., ter Horst, J.R., Verbeek, M., 2005. Survival, look-ahead bias, and persis-
tence in hedge fund performance. Journal of Financial and Quantitative Analysis
40 (3), 493-517.
Bekaert, G., Urias, M.S., 1996. Diversification, integration and emerging market
closed-ended funds. Journal of Finance 51 (2), 835-870.
Boyson, N.M., Stahel, C.W., Stulz, R.M., 2008. Why do hedge funds worst returns
cluster? Common liquidity shocks vs. contagion. Fisher College of Business Work-
ing Paper No. 2008-03-007.
Brown, S.J., Goetzmann, W.N., Park, J., 2001. Careers and survival: competition and
risks in the hedge fund and CTA industry. Journal of Finance 56 (5), 1869-1886.
Brunnermeier, M.K., Pedersen, L.H., 2009. Market liquidity and funding liquidity.
Review of Financial Studies 22 (6), 2201-2238.
35
ECB
Working Paper Series No 1112
November 2009
Butler, K.C., Joaquin, D.C., 2002. Are the gains from international portfolio diversifi-
cation exaggerated? The influence of downside risk in bear markets. Journal of In-
ternational Money and Finance 21 (7), 981-1011.
Campbell, J.Y., Hilscher, J., Szilagyi, J., 2008. In search of distress risk. Journal of
Finance 63 (6), 2899-2939.
Casey, Quirk & Associates and the Bank of New York, 2006. Institutional demand for
hedge funds 2: a global perspective. White Paper, October.
Chan, N., Getmansky, M., Haas, S.M., Lo, A.W., 2005. Systemic risk and hedge
funds. NBER Working Paper No. 11200.
Chen, Q., Goldstein, I., Jiang, W., 2007. Payoff complementarities and financial fra-
gility: Evidence from mutual fund outflows. Working Paper.
Chordia, T., 1996. The structure of mutual fund charges. Journal of Financial Eco-
nomics 41 (1), 3-39.
Chowdry, B., Nanda, V., 1998. Leverage and market stability: the role of margin rules
and price limits. Journal of Business 71 (2), 179-210.
Cifuentes, R.G., Ferrucci G., Shin, H.S., 2005. Liquidity risk and contagion. Working
Paper No. 264, Bank of England.
De Roon, F.A., Nijman, T.E., Werker, B.J.M., 2001. Testing for mean-variance span-
ning with short sales constraints and transaction costs: the case of emerging mar-
kets. Journal of Finance 56 (2), 721-742.
De Santis, G., Gerard, B., 1997. International asset pricing and portfolio diversifica-
tion with time-varying risk. Journal of Finance 52 (4), 1881-1912.
Driessen, J., Laeven, L., 2007. International portfolio diversification benefits: cross-
country evidence from a local perspective. Journal of Banking and Finance 31 (6),
1693-1712.
Dungey, M., Fry, R., Gonzlez-Hermosillo, B., Martin, V.L., 2005. Empirical model-
ling of contagion: A review of methodologies. Quantitative Finance 5 (1), 9-24.
36
ECB
Working Paper Series No 1112
November 2009
Edelen, R.M., 1999. Investor flows and the assessed performance of open-end mutual
funds. Journal of Financial Economics 53 (3), 439-466.
Eichengreen, B., Rose, A., Wyplosz, C., 1996. Contagious currency crises: first tests.
Scandinavian Journal of Economics 98 (4), 463-484.
Ewerhart, C., Valla, N., 2007. Forced portfolio liquidation, Notes dEtudes et de Re-
cherche, NER-R # 179, Banque de France.
Financial Stability Forum, 2007. Update of the FSF report of highly leveraged institu-
tions, BIS.
Forbes, K., Rigobon, R., 2002. No contagion, only interdependence: measuring stock
market co-movements. Journal of Finance 43 (5), 2223-2261.
Fung, W., Hsieh, D., 1997. Survivorship bias and investment style in the returns of
CTAs. Journal of Portfolio Management 24 (1), 30-41.
Getmansky, M., 2005. The life cycle of hedge funds: Fund flows, size and perform-
ance. Working Paper.
Giesecke, K., 2004. Correlated default with incomplete information. Journal of Bank-
ing and Finance 28 (7), 1521-1545.
Goetzmann, W.N., Ingersoll, J.E., Ross, S.A., 2003. High-water marks and hedge
fund management contracts. Journal of Finance 58 (4), 1685-1717.
Greenwich Associates, 2007. Global hedge fund and prime brokerage trends. May
2007.
Gregoriou, N.G., 2002. Hedge fund survival lifetimes. Journal of Asset Management
3 (3), 237-252.
Harvey, C.R., 1995. Predictable risk and returns in emerging markets. Review of Fi-
nancial Studies 8 (3), 773-816.
International Monetary Fund, 2008. Global Financial Stability Report. October 2008.
37
ECB
Working Paper Series No 1112
November 2009
Kolokolova, O., 2009. Birth and death of family hedge funds: The determinants.
Working Paper.
Liang, B., 2000. Hedge funds: the living and the dead. Journal of Financial and Quan-
titative Analysis 35 (3), 309-326.
Liu, X., Mello, A.S., 2009. The fragile capital structure of hedge funds and the limits
to arbitrage. Working Paper.
Lo, A.W., 2008. Hedge Funds, Systemic Risk, and the Financial Crisis of 2007-2008.
Working Paper.
Longin; F.; Solnik, B., 2001. Extreme correlation of international equity markets.
Journal of Finance 56 (2), 649-676.
Lunde, A., Timmermann, A.G., Blake, D., 1999. The hazards of mutual fund under-
performance: A Cox regression analysis. Journal of Empirical Finance 6 (2), 121-
152.
Nanda, V., Narayanan, M.P., Warther, V.A., 2000. Liquidity, investment ability, and
mutual fund structure. Journal of Financial Economics 57 (3), 417-443.
Panageas, S., Westerfield, M., 2008. High water marks: high risks appetites? Convex
compensation, long horizons and portfolio choice. Journal of Finance, forthcom-
ing.
Pericoli, M., Sbracia, M., 2003. A primer on financial contagion. Journal of Economic
Surveys 17 (4), 571-608.
Pesaran, M.H., Pick, A., 2007. Econometric issues in the analysis of contagion. Jour-
nal of Economic Dynamics and Control 31 (4), 1245-1277.
38
ECB
Working Paper Series No 1112
November 2009
Appendix
Table 1: Number of Hedge Funds, Entries, Exits and Failures
For each year of the sample period, this table shows the number of existing funds, new entries into the
TASS database, exits out of the database, fund failures, the attrition and failure rates in percent, the
mean and the standard deviation of monthly hedge fund returns in percent and the mean of assets under
management in million USD, separately for funds classified as Alive and Liquidated. The number
of existing funds refers to the number of funds classified as Alive at the end of each year. The attri-
tion (failure) rate is the ratio of exits (failures) to the number of existing funds. The asterisk indicates
that for 2008 data have been included only until end of May.
Alive Funds Liquidated Funds
Year
Existing
Funds
New
Entries Exits
Fail-
ures
Attrition
Rate in
%
Failure
Rate in %
Mean
Monthly
Return
Std.
Dev.
Monthly
Return
Mean
AUM
in mn
USD
Mean
Monthly
Return
Std.
Dev.
Monthly
Return
Mean
AUM
in mn
USD
1994 524 170 6 3 1.15 0.57 0.11 4.95 74.85 -0.05 4.62 61.63
1995 660 162 26 18 3.94 2.73 1.60 4.60 61.69 0.90 5.08 46.61
1996 790 216 86 37 10.89 4.68 1.82 4.68 72.81 0.99 5.39 43.02
1997 1007 271 54 35 5.36 3.48 1.59 4.92 88.78 1.05 5.60 53.72
1998 1148 245 104 73 9.06 6.36 0.36 6.54 89.48 0.23 6.86 67.15
1999 1350 320 118 70 8.74 5.19 2.11 5.85 84.44 1.27 6.01 48.17
2000 1532 328 146 64 9.53 4.18 1.19 5.94 96.53 0.41 6.69 42.88
2001 1733 359 158 69 9.12 3.98 0.83 4.50 100.49 0.08 4.86 48.48
2002 1983 418 168 92 8.47 4.64 0.45 4.11 102.69 -0.07 3.91 46.68
2003 2268 447 162 107 7.14 4.72 1.51 3.62 118.62 0.88 3.33 50.43
2004 2589 550 229 133 8.85 5.14 0.83 3.08 154.16 0.27 2.85 64.61
2005 2736 472 325 163 11.88 5.96 0.91 3.21 167.13 0.38 3.04 72.22
2006 2930 570 376 144 12.83 4.91 1.05 3.21 183.72 0.32 2.98 82.37
2007 3140 663 452 92 14.39 2.93 0.91 3.50 208.14 0.02 3.16 88.81
*2008 2750 8 243 44 8.84 1.60 -0.04 4.53 222.09 -1.41 3.83 58.12
39
ECB
Working Paper Series No 1112
November 2009
Table 2: Summary statistics of hedge fund variables
This table shows the summary statistics of various hedge fund characteristics that are used as explana-
tory variables in our estimations. The summary statistics are computed using the data after filtering
over the sample period from January 1994 to May 2008.
Variable Mean SD
25th
Percentile Median
75th
Percentile
Return per month in % 0.79 4.49 -0.69 0.70 2.16
Age in years 4.51 3.77 1.75 3.42 6.17
AUM in mn USD 111.39 222.13 9.20 32.10 105.54
Monthly capital flows in % of AUM 1.88 11.68 -0.86 0.07 2.69
Leverage (% of funds) 66.95
Management fee in % 1.47 0.69 1.00 1.50 2.00
Incentive fee in % 16.59 7.19 15.00 20.00 20.00
High watermark (% of funds) 61.67
Lockup period in months 3.06 5.83 0.00 0.00 3.00
Redemption frequency in months 2.31 2.63 1.00 1.00 3.00
Redemption notice period in months 1.14 0.90 0.47 1.00 1.50
Payout period in months 0.49 0.62 0.00 0.33 1.00
Minimum investment in mn USD 0.77 2.10 0.10 0.32 1.00
Month 12 (% of obs.) 8.37
Invested in derivatives (% of funds) 61.36
Redemption within 0.38 0.18 0.24 0.36 0.48
Redemption across 0.37 0.13 0.28 0.37 0.45
Failure rate within 0.19 0.40 0.00 0.00 0.23
Failure rate across 0.39 0.25 0.23 0.34 0.52
40
ECB
Working Paper Series No 1112
November 2009
Table 3: Estimation Results Risk Spillover
This table reports the coefficient estimates of a logit model for hedge fund failure of equation (4), re-
ferred to as the Baseline Model, of equation (5a), which models both channels of risk spillover (inves-
tor redemptions and failures of other funds) separately, and of equation (5b), which models both chan-
nels of risk spillover simultaneously. The dependent variable takes on the value 1 in the month where
the hedge fund fails, and is 0 in all prior months. To facilitate comparisons of the coefficients, all non-
indicator explanatory variables have been standardized to have a zero mean and a standard deviation of
one. To account for fixed effects associated with the calendar year and the investment style, indicator
variables are included in each of the models. The results are not reported here, but are available upon
request. The sample period extends from January 1994 to May 2008. The estimation results are ob-
tained by maximum likelihood. Statistical significance at the 1%, 5%, and 10% levels is denoted by
***, **, and *, respectively.
Risk Spillover Models
(1) (2) (3) (4)
Regressor
Baseline
Model
Redemption
Ratio
Failure
Rate Combined
Age -0.073* -0.072* -0.073* -0.072*
Return -0.279*** -0.274*** -0.273*** -0.267***
Return (t-1) -0.151*** -0.145*** -0.153*** -0.146***
Return (t-2) -0.131*** -0.134*** -0.129*** -0.134***
log(AUM (t-1)) -0.688*** -0.688*** -0.688*** -0.688***
Flows -0.315*** -0.312*** -0.312*** -0.310***
Flows (t-1) -0.330*** -0.330*** -0.328*** -0.329***
Flows (t-2) -0.397*** -0.393*** -0.397*** -0.393***
Leverage 0.006 0.004 0.007 0.005
Management fee 0.026 0.026 0.025 0.025
Incentive fee 0.091** 0.091** 0.091** 0.091**
High watermark -0.371*** -0.375*** -0.372*** -0.375***
Lockup period -0.093** -0.096** -0.094** -0.096**
Redemption frequency 0.051 0.052 0.052 0.053
Redemption notice period -0.301*** -0.302*** -0.303*** -0.304***
Payout period -0.121*** -0.122*** -0.122*** -0.123***
Minimum investment 0.144 0.153 0.146 0.154
Month 12 0.885*** 0.928*** 0.878*** 0.913***
Invested in derivatives 0.269*** 0.271*** 0.271*** 0.272***
Spillover variables
Redemption within (t-1) 0.001 -0.020
Redemption within (t-2) 0.104*** 0.108***
Redemption across (t-1) -0.131*** -0.140***
Redemption across (t-2) -0.107** -0.092*
Failure within (t-1) 0.053** 0.051*
Failure within (t-2) 0.054** 0.050*
Failure across (t-1) 0.036 0.048
Failure across (t-2) -0.030 -0.002
Constant -8.438*** -8.383*** -8.394*** -8.279***
Number of Obs. 270747 270747 270747 270747
LR chi2 (df) 1358.23 (43) 1380.98 (47) 1368.53 (47) 1390.92 (51)
Prob > Chi2 0.0000 0.0000 0.0000 0.0000
Pseudo R-Squared 0.1028 0.1045 0.1036 0.1053
Log-Likelihood -5925.82 -5914.44 -5920.67 -5909.47
LR-test value (df) - - - 32.69 (8)
41
ECB
Working Paper Series No 1112
November 2009
T
a
b
l
e
4
:
E
s
t
i
m
a
t
i
o
n
R
e
s
u
l
t
s
D
i
v
e
r
s
i
f
i
c
a
t
i
o
n
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
t
h
e
c
o
e
f
f
i
c
i
e
n
t
e
s
t
i
m
a
t
e
s
o
f
a
l
o
g
i
t
m
o
d
e
l
f
o
r
h
e
d
g
e
f
u
n
d
f
a
i
l
u
r
e
o
f
e
q
u
a
t
i
o
n
(
6
a
)
,
w
h
i
c
h
a
n
a
l
y
s
e
s
t
h
e
e
f
f
e
c
t
o
f
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
i
n
t
e
r
m
s
o
f
g
e
o
g
r
a
p
h
y
a
n
d
a
s
-
s
e
t
t
y
p
e
,
o
f
e
q
u
a
t
i
o
n
(
6
b
)
,
w
h
i
c
h
i
n
v
e
s
t
i
g
a
t
e
s
t
h
e
i
m
p
a
c
t
o
f
b
e
i
n
g
i
n
v
e
s
t
e
d
i
n
e
q
u
i
t
i
e
s
a
n
d
/
o
r
b
o
n
d
s
,
a
n
d
o
f
e
q
u
a
t
i
o
n
(
6
c
)
,
w
h
i
c
h
a
n
a
l
y
s
e
s
t
h
e
i
m
p
a
c
t
o
f
d
i
f
f
e
r
e
n
t
d
e
g
r
e
e
s
o
f
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
r
a
n
g
i
n
g
f
r
o
m
N
o
t
d
i
v
e
r
s
i
f
i
e
d
a
t
a
l
l
t
o
D
i
v
e
r
s
i
f
i
e
d
i
n
a
s
s
e
t
s
&
g
e
o
g
r
a
p
h
i
c
a
l
l
y
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
t
a
k
e
s
o
n
t
h
e
v
a
l
u
e
1
i
n
t
h
e
m
o
n
t
h
w
h
e
r
e
t
h
e
h
e
d
g
e
f
u
n
d
f
a
i
l
s
,
a
n
d
i
s
0
i
n
a
l
l
p
r
i
o
r
m
o
n
t
h
s
.
T
h
e
r
e
g
r
e
s
s
o
r
s
o
f
t
h
e
B
a
s
e
l
i
n
e
M
o
d
e
l
a
r
e
i
n
c
l
u
d
e
d
i
n
t
h
e
e
s
t
i
m
a
t
i
o
n
o
f
a
l
l
m
o
d
e
l
s
,
b
u
t
a
r
e
n
o
t
r
e
p
o
r
t
e
d
f
o
r
b
r
e
v
i
t
y
.
T
o
a
c
c
o
u
n
t
f
o
r
f
i
x
e
d
e
f
f
e
c
t
s
a
s
s
o
c
i
a
t
e
d
w
i
t
h
t
h
e
c
a
l
e
n
d
a
r
y
e
a
r
a
n
d
t
h
e
i
n
v
e
s
t
m
e
n
t
s
t
y
l
e
,
i
n
d
i
c
a
t
o
r
v
a
r
i
a
b
l
e
s
a
r
e
i
n
c
l
u
d
e
d
i
n
e
a
c
h
o
f
t
h
e
m
o
d
e
l
s
.
T
h
e
r
e
s
u
l
t
s
a
r
e
n
o
t
r
e
p
o
r
t
e
d
h
e
r
e
,
b
u
t
a
r
e
a
v
a
i
l
-
a
b
l
e
u
p
o
n
r
e
q
u
e
s
t
.
A
l
s
o
n
o
t
r
e
p
o
r
t
e
d
a
r
e
t
h
e
c
o
e
f
f
i
c
i
e
n
t
e
s
t
i
m
a
t
e
s
o
f
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
a
c
c
o
u
n
t
i
n
g
f
o
r
t
h
e
c
o
m
m
o
d
i
t
i
e
s
,
F
X
a
n
d
r
e
a
l
e
s
t
a
t
e
m
a
r
k
e
t
s
a
s
t
h
e
y
o
n
l
y
a
c
c
o
u
n
t
f
o
r
a
v
e
r
y
s
m
a
l
l
p
r
o
p
o
r
t
i
o
n
o
f
f
u
n
d
s
.
T
h
e
s
a
m
p
l
e
p
e
r
i
o
d
e
x
t
e
n
d
s
f
r
o
m
J
a
n
u
a
r
y
1
9
9
4
t
o
M
a
y
2
0
0
8
.
T
h
e
e
s
t
i
m
a
t
i
o
n
r
e
s
u
l
t
s
a
r
e
o
b
t
a
i
n
e
d
b
y
m
a
x
i
m
u
m
l
i
k
e
l
i
h
o
o
d
.
S
t
a
t
i
s
t
i
c
a
l
s
i
g
n
i
f
i
c
a
n
c
e
a
t
t
h
e
1
%
,
5
%
,
a
n
d
1
0
%
l
e
v
e
l
s
i
s
d
e
n
o
t
e
d
b
y
*
*
*
,
*
*
,
a
n
d
*
,
r
e
s
p
e
c
t
i
v
e
l
y
.
R
e
f
e
r
e
n
c
e
c
a
t
e
g
o
r
y
R
e
f
e
r
e
n
c
e
c
a
t
e
g
o
r
y
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
(
7
)
(
8
)
(
9
)
R
e
g
r
e
s
s
o
r
G
e
o
g
r
a
p
h
i
c
a
l
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
A
s
s
e
t
s
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
I
n
v
e
s
t
m
e
n
t
s
i
n
E
q
u
i
t
y
o
n
l
y
I
n
v
e
s
t
m
e
n
t
s
i
n
B
o
n
d
s
o
n
l
y
I
n
v
e
s
t
m
e
n
t
s
i
n
B
o
n
d
s
&
E
q
u
i
-
t
i
e
s
D
i
v
e
r
s
i
f
i
e
d
i
n
a
s
s
e
t
s
&
g
e
o
g
r
a
p
h
i
c
a
l
l
y
N
o
t
d
i
v
e
r
s
i
f
i
e
d
a
t
a
l
l
D
i
v
e
r
s
i
f
i
e
d
i
n
a
s
s
e
t
s
o
n
l
y
D
i
v
e
r
s
i
f
i
e
d
o
n
l
y
g
e
o
g
r
a
p
h
i
c
a
l
l
y
D
i
v
e
r
s
i
f
i
c
a
t
i
o
n
v
a
r
i
a
b
l
e
s
G
e
o
g
r
a
p
h
i
c
a
l
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
-
0
.
1
8
5
*
*
A
s
s
e
t
s
t
y
p
e
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
-
0
.
2
1
2
*
*
*
A
s
s
e
t
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
v
a
r
i
a
b
l
e
s
I
n
v
e
s
t
.
i
n
e
q
u
i
t
i
e
s
o
n
l
y
-
-
0
.
0
7
3
0
.
2
5
4
*
*
*
I
n
v
e
s
t
.
i
n
b
o
n
d
s
o
n
l
y
0
.
0
7
3
-
0
.
3
2
6
*
I
n
v
e
s
t
.
i
n
b
o
n
d
s
&
e
q
u
i
t
i
e
s
-
0
.
2
5
4
*
*
*
-
0
.
3
2
6
*
-
C
o
m
b
i
n
e
d
d
i
v
e
r
s
i
f
i
c
a
t
i
o
n
v
a
r
i
a
b
l
e
s
D
i
v
e
r
s
i
f
i
e
d
i
n
a
s
s
e
t
s
&
g
e
o
g
r
a
p
h
i
c
a
l
l
y
-
-
0
.
3
4
5
*
*
*
-
0
.
1
7
6
*
-
0
.
2
2
0
*
N
o
t
d
i
v
e
r
s
i
f
i
e
d
a
t
a
l
l
0
.
3
4
5
*
*
*
-
0
.
1
6
9
*
0
.
1
2
5
D
i
v
e
r
s
i
f
i
e
d
i
n
a
s
s
e
t
s
o
n
l
y
0
.
1
7
6
*
-
0
.
1
6
9
*
-
-
0
.
0
4
4
D
i
v
e
r
s
i
f
i
e
d
g
e
o
g
r
a
p
h
i
c
a
l
l
y
o
n
l
y
0
.
2
2
0
*
-
0
.
1
2
5
0
.
0
4
4
-
N
u
m
b
e
r
o
f
O
b
s
.
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
2
7
0
7
4
7
L
R
c
h
i
2
(
d
f
)
1
3
6
4
.
2
7
(
4
4
)
1
3
6
6
.
5
2
(
4
4
)
1
3
6
8
.
7
5
(
4
7
)
1
3
6
8
.
7
5
(
4
7
)
1
3
6
8
.
7
5
(
4
7
)
1
3
7
0
.
8
1
(
4
6
)
1
3
7
0
.
8
1
(
4
6
)
1
3
7
0
.
8
1
(
4
6
)
1
3
7
0
.
8
1
(
4
6
)
P
r
o
b
>
C
h
i
2
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
0
0
0
P
s
e
u
d
o
R
-
S
q
u
a
r
e
d
0
.
1
0
3
3
0
.
1
0
3
4
0
.
1
0
3
6
0
.
1
0
3
6
0
.
1
0
3
6
0
.
1
0
3
8
0
.
1
0
3
8
0
.
1
0
3
8
0
.
1
0
3
8
L
o
g
-
L
i
k
e
l
i
h
o
o
d
-
5
9
2
2
.
8
0
-
5
9
2
1
.
6
7
-
5
9
2
0
.
5
6
-
5
9
2
0
.
5
6
-
5
9
2
0
.
5
6
-
5
9
1
9
.
5
2
9
4
-
5
9
1
9
.
5
2
9
4
-
5
9
1
9
.
5
2
9
4
-
5
9
1
9
.
5
2
9
4
L
R
-
t
e
s
t
v
a
l
u
e
(
d
f
)
-
-
-
-
-
-
1
2
.
5
8
0
(
3
)
-
-
42
ECB
Working Paper Series No 1112
November 2009
Table 5: Estimation Results Interaction of Risk Spillover and Diversification
This table reports the coefficient estimates of a logit model for hedge fund failure of equation (7),
which analyses the interaction of risk spillover and diversification. The dependent variable takes on the
value 1 in the month where the hedge fund fails, and is 0 in all prior months. The regressors of the
Baseline Model are included in the estimation, but are not reported for brevity. To account for fixed
effects associated with the calendar year and the investment style, indicator variables are included. The
results are not reported here, but are available upon request. The sample period extends from January
1994 to May 2008. The estimation results are obtained by maximum likelihood. Statistical significance
at the 1%, 5%, and 10% levels is denoted by ***, **, and *, respectively.
Regressor Coefficient
Diversification variable
Diversified in assets & geographically -0.215**
Spillover variables
Redemption within (t-1) -0.049
Redemption within (t-2) 0.152***
Redemption across (t-1) -0.218***
Redemption across (t-2) -0.055
Failure within (t-1) 0.073**
Failure within (t-2) 0.082**
Failure across (t-1) 0.039
Failure across (t-2) 0.019
Interaction variables
Diversified x redemption within (t-1) 0.085
Diversified x redemption within (t-2) -0.153*
Diversified x redemption across (t-1) 0.245**
Diversified x redemption across (t-2) -0.115
Diversified x failure within (t-1) -0.059
Diversified x failure within (t-2) -0.091
Diversified x failure across (t-1) 0.034
Diversified x failure across (t-2) -0.061
Number of Obs. 270747
LR chi2 (df) 1416.76 (60)
Prob > Chi2 0.0000
Pseudo R-Squared 0.1072
Log-Likelihood -5896.5541
LR-test value (df) 16.841 (8)
43
ECB
Working Paper Series No 1112
November 2009
T
a
b
l
e
6
:
E
s
t
i
m
a
t
i
o
n
R
e
s
u
l
t
s
R
o
b
u
s
t
n
e
s
s
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
t
h
e
c
o
e
f
f
i
c
i
e
n
t
e
s
t
i
m
a
t
e
s
o
f
d
i
f
f
e
r
e
n
t
m
o
d
e
l
s
f
o
r
h
e
d
g
e
f
u
n
d
f
a
i
l
u
r
e
t
o
a
n
a
l
y
s
e
t
h
e
r
o
b
u
s
t
n
e
s
s
o
f
o
u
r
r
e
s
u
l
t
s
.
F
o
r
c
o
n
v
e
n
i
e
n
c
e
,
w
e
o
n
l
y
r
e
p
o
r
t
t
h
e
c
o
e
f
f
i
c
i
e
n
t
s
o
f
t
h
e
v
a
r
i
a
b
l
e
s
c
a
p
t
u
r
i
n
g
r
i
s
k
s
p
i
l
l
o
v
e
r
a
n
d
s
u
p
p
r
e
s
s
t
h
e
r
e
p
o
r
t
i
n
g
o
f
t
h
e
c
o
e
f
f
i
c
i
e
n
t
s
o
f
o
t
h
e
r
v
a
r
i
a
b
l
e
s
.
I
n
e
a
c
h
o
f
t
h
e
r
o
w
s
1
t
o
4
t
h
e
e
s
t
i
m
a
t
i
o
n
s
a
r
e
b
a
s
e
d
o
n
e
q
u
a
t
i
o
n
(
5
b
)
,
u
s
i
n
g
a
d
i
f
f
e
r
e
n
t
m
o
d
e
l
(
l
o
g
i
t
v
s
.
C
o
x
p
r
o
p
o
r
t
i
o
n
a
l
h
a
z
a
r
d
s
r
a
t
e
)
a
n
d
/
o
r
a
d
i
f
f
e
r
e
n
t
s
p
e
c
i
f
i
c
a
t
i
o
n
o
f
f
u
n
d
f
a
i
l
u
r
e
.
R
o
w
5
i
s
b
a
s
e
d
o
n
a
m
o
d
i
f
i
e
d
v
e
r
s
i
o
n
o
f
e
q
u
a
t
i
o
n
(
5
b
)
u
s
i
n
g
r
i
s
k
-
a
d
j
u
s
t
e
d
i
n
s
t
e
a
d
o
f
r
a
w
r
e
t
u
r
n
s
a
s
a
c
o
n
t
r
o
l
v
a
r
i
a
b
l
e
c
a
p
t
u
r
i
n
g
h
e
d
g
e
f
u
n
d
p
e
r
f
o
r
m
a
n
c
e
.
T
o
f
a
c
i
l
i
t
a
t
e
c
o
m
p
a
r
i
s
o
n
s
,
w
e
a
l
s
o
r
e
p
o
r
t
t
h
e
e
s
t
i
m
a
t
i
o
n
r
e
s
u
l
t
s
f
r
o
m
T
a
b
l
e
3
,
w
h
i
c
h
a
r
e
b
a
s
e
d
o
n
a
l
o
g
i
t
m
o
d
e
l
w
h
e
r
e
G
r
a
v
e
y
a
r
d
f
u
n
d
s
n
o
t
c
l
a
s
s
i
f
i
e
d
a
s
L
i
q
u
i
d
a
t
e
d
a
r
e
t
r
e
a
t
e
d
a
s
c
e
n
s
o
r
e
d
a
t
t
h
e
d
a
t
e
o
f
l
a
s
t
r
e
p
o
r
t
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
t
a
k
e
s
o
n
t
h
e
v
a
l
u
e
1
i
n
t
h
e
m
o
n
t
h
w
h
e
r
e
t
h
e
h
e
d
g
e
f
u
n
d
f
a
i
l
s
,
a
n
d
i
s
0
i
n
a
l
l
p
r
i
o
r
m
o
n
t
h
s
.
T
h
e
r
e
g
r
e
s
s
o
r
s
o
f
t
h
e
B
a
s
e
l
i
n
e
M
o
d
e
l
a
r
e
i
n
c
l
u
d
e
d
i
n
t
h
e
e
s
t
i
m
a
t
i
o
n
o
f
a
l
l
m
o
d
e
l
s
,
b
u
t
a
r
e
n
o
t
r
e
p
o
r
t
e
d
f
o
r
b
r
e
v
i
t
y
.
T
o
a
c
c
o
u
n
t
f
o
r
f
i
x
e
d
e
f
f
e
c
t
s
a
s
s
o
c
i
a
t
e
d
w
i
t
h
t
h
e
c
a
l
e
n
d
a
r
y
e
a
r
a
n
d
t
h
e
i
n
v
e
s
t
m
e
n
t
s
t
y
l
e
,
i
n
d
i
c
a
t
o
r
v
a
r
i
a
b
l
e
s
a
r
e
i
n
c
l
u
d
e
d
i
n
e
a
c
h
o
f
t
h
e
m
o
d
e
l
s
.
T
h
e
r
e
s
u
l
t
s
a
r
e
n
o
t
r
e
p
o
r
t
e
d
h
e
r
e
,
b
u
t
a
r
e
a
v
a
i
l
a
b
l
e
u
p
o
n
r
e
q
u
e
s
t
.
T
h
e
s
a
m
p
l
e
p
e
r
i
o
d
e
x
t
e
n
d
s
f
r
o
m
J
a
n
u
a
r
y
1
9
9
4
t
o
M
a
y
2
0
0
8
.
T
h
e
e
s
t
i
m
a
t
i
o
n
r
e
s
u
l
t
s
a
r
e
o
b
t
a
i
n
e
d
b
y
m
a
x
i
m
u
m
l
i
k
e
l
i
-
h
o
o
d
.
S
t
a
t
i
s
t
i
c
a
l
s
i
g
n
i
f
i
c
a
n
c
e
a
t
t
h
e
1
%
,
5
%
,
a
n
d
1
0
%
l
e
v
e
l
s
i
s
d
e
n
o
t
e
d
b
y
*
*
*
,
*
*
,
a
n
d
*
,
r
e
s
p
e
c
t
i
v
e
l
y
.
R
o
w
T
y
p
e
o
f
r
o
b
u
s
t
n
e
s
s
R
e
d
e
m
p
t
i
o
n
w
i
t
h
i
n
(
t
-
1
)
R
e
d
e
m
p
t
i
o
n
w
i
t
h
i
n
(
t
-
2
)
R
e
d
e
m
p
t
i
o
n
a
c
r
o
s
s
(
t
-
1
)
R
e
d
e
m
p
t
i
o
n
a
c
r
o
s
s
(
t
-
2
)
F
a
i
l
u
r
e
w
i
t
h
i
n
(
t
-
1
)
F
a
i
l
u
r
e
w
i
t
h
i
n
(
t
-
2
)
F
a
i
l
u
r
e
a
c
r
o
s
s
(
t
-
1
)
F
a
i
l
u
r
e
a
c
r
o
s
s
(
t
-
2
)
R
e
s
u
l
t
f
r
o
m
T
a
b
l
e
3
-
0
.
0
2
0
.
1
0
8
*
*
*
-
0
.
1
4
0
*
*
*
-
0
.
0
9
2
*
0
.
0
5
1
*
0
.
0
5
0
*
0
.
0
4
8
-
0
.
0
0
2
L
o
g
i
t
m
o
d
e
l
1
B
a
q
u
e
r
o
e
t
a
l
.
(
2
0
0
5
)
f
l
o
w
-
p
r
o
c
e
d
u
r
e
-
0
.
0
1
0
0
.
0
8
0
*
*
*
-
0
.
1
4
6
*
*
*
0
.
0
0
0
0
.
0
4
3
*
*
0
.
0
1
2
0
.
0
1
3
-
0
.
0
7
4
*
*
2
O
n
l
y
A
l
i
v
e
a
n
d
L
i
q
u
i
d
a
t
e
d
f
u
n
d
s
-
0
.
0
1
6
0
.
1
1
0
*
*
*
-
0
.
1
4
3
*
*
*
-
0
.
0
7
6
0
.
0
4
7
*
0
.
0
4
7
*
0
.
0
5
6
0
.
0
0
0
C
o
x
p
r
o
p
o
r
t
i
o
n
a
l
h
a
z
a
r
d
s
r
a
t
e
m
o
d
e
l
3
B
a
q
u
e
r
o
e
t
a
l
.
(
2
0
0
5
)
f
l
o
w
-
p
r
o
c
e
d
u
r
e
-
0
.
0
1
1
0
.
0
7
7
*
*
-
0
.
1
3
9
*
*
*
-
0
.
0
0
1
0
.
0
3
6
*
0
.
0
1
1
0
.
0
2
-
0
.
0
7
4
*
*
4
O
n
l
y
A
l
i
v
e
a
n
d
L
i
q
u
i
d
a
t
e
d
f
u
n
d
s
-
0
.
0
1
8
0
.
1
0
5
*
*
*
-
0
.
1
3
8
*
*
*
-
0
.
0
9
4
*
0
.
0
4
2
0
.
0
4
7
*
0
.
0
5
2
-
0
.
0
0
2
L
o
g
i
t
m
o
d
e
l
u
s
i
n
g
r
i
s
k
-
a
d
j
u
s
t
e
d
r
e
t
u
r
n
s
5
6
-
m
o
n
t
h
r
o
l
l
i
n
g
S
h
a
r
p
e
r
a
t
i
o
-
0
.
0
4
6
0
.
0
8
3
*
*
-
0
.
1
7
7
*
*
*
-
0
.
1
1
4
*
*
0
.
0
5
6
*
*
0
.
0
5
8
*
*
0
.
0
5
4
0
.
0
0
2
44
ECB
Working Paper Series No 1112
November 2009
Table 7: Marginal Effects
This table reports the monthly failure probability of an exemplary hedge fund based on equation (5b),
which is referred to as the Baseline Failure Probability, where all explanatory variables have been set
to the sample mean, the failure probability which results from a one standard deviation increase in the
corresponding regressor all other variables left unchanged, the marginal effect which is computed as
the difference of the two failure probabilities and the 95% confidence interval of the marginal effect.
Dummy variables, labelled with an asterisk, are set to 0 in the Baseline case and 1 in the Stress case.
As our models include yearly fixed effects, we have to specify a year for which the failure probabilities
in this table are computed. Therefore, we assume that the exemplary hedge fund operates in 2007.
Only variables which are significant at the 10% level are reported.
Regressor
Baseline
Failure Probability
Failure Probability
(1 Std. Dev.
Increase in
Regressor)
Marginal
Effect
95% Conf. Interval
of Marginal Effect
Variables increasing the failure probability
Invested in derivatives* 0.2005% 0.2629% 0.0624% 0.0118% 0.2466%
Redemption within (t-2) 0.2005% 0.2229% 0.0225% 0.0024% 0.0991%
Incentive fee 0.2005% 0.2193% 0.0188% 0.0008% 0.0910%
Failure within (t-1) 0.2005% 0.2108% 0.0104% -0.0001% 0.0527%
Failure within (t-2) 0.2005% 0.2106% 0.0102% -0.0002% 0.0521%
Variables reducing the failure probability
log(AUM(t-1)) 0.2005% 0.1020% -0.0984% -0.0427% -0.2243%
Flows (t-2) 0.2005% 0.1360% -0.0645% -0.0304% -0.1307%
High watermark* 0.2005% 0.1379% -0.0625% -0.0338% -0.0944%
Flows (t-1) 0.2005% 0.1468% -0.0536% -0.0262% -0.1029%
Redemption notice period 0.2005% 0.1481% -0.0523% -0.0267% -0.0926%
Flows 0.2005% 0.1506% -0.0498% -0.0246% -0.0939%
Return 0.2005% 0.1539% -0.0466% -0.0221% -0.0944%
Return (t-1) 0.2005% 0.1736% -0.0269% -0.0146% -0.0426%
Redemption across (t-1) 0.2005% 0.1745% -0.0260% -0.0173% -0.0188%
Return (t-2) 0.2005% 0.1756% -0.0249% -0.0139% -0.0367%
Payout period 0.2005% 0.1774% -0.0231% -0.0155% -0.0162%
Lockup period 0.2005% 0.1820% -0.0184% -0.0135% -0.0058%
Redemption across (t-2) 0.2005% 0.1831% -0.0173% -0.0140% 0.0039%
Age 0.2005% 0.1868% -0.0137% -0.0111% 0.0023%
45
ECB
Working Paper Series No 1112
November 2009
European Central Bank Working Paper Series
For a complete list of Working Papers published by the ECB, please visit the ECBs website
(http://www.ecb.europa.eu).
1077 The reception of public signals in financial markets what if central bank communication becomes stale?
by M. Ehrmann and D. Sondermann, August 2009.
1078 On the real effects of private equity investment: evidence from new business creation by A. Popov and
P. Roosenboom, August 2009.
1079 EMU and European government bond market integration by P. Abad and H. Chuli, and M. Gmez-Puig,
August 2009.
1080 Productivity and job flows: heterogeneity of new hires and continuing jobs in the business cycle by J. Kilponen
and J. Vanhala, August 2009.
1081 Liquidity premia in German government bonds by J. W. Ejsing and J. Sihvonen, August 2009.
1082 Disagreement among forecasters in G7 countries by J. Dovern, U. Fritsche and J. Slacalek, August 2009.
1083 Evaluating microfoundations for aggregate price rigidities: evidence from matched firm-level data on product
prices and unit labor cost by M. Carlsson and O. Nordstrm Skans, August 2009.
1084 How are firms wages and prices linked: survey evidence in Europe by M. Druant, S. Fabiani, G. Kezdi,
A. Lamo, F. Martins and R. Sabbatini, August 2009.
1085 An empirical study on the decoupling movements between corporate bond and CDS spreads
by I. Alexopoulou, M. Andersson and O. M. Georgescu, August 2009.
1086 Euro area money demand: empirical evidence on the role of equity and labour markets by G. J. de Bondt,
September 2009.
1087 Modelling global trade flows: results from a GVAR model by M. Bussire, A. Chudik and G. Sestieri,
September 2009.
1088 Inflation perceptions and expectations in the euro area: the role of news by C. Badarinza and M. Buchmann,
September 2009.
1089 The effects of monetary policy on unemployment dynamics under model uncertainty: evidence from the US
and the euro area by C. Altavilla and M. Ciccarelli, September 2009.
1090 New Keynesian versus old Keynesian government spending multipliers by J. F. Cogan, T. Cwik, J. B. Taylor
and V. Wieland, September 2009.
1091 Money talks by M. Hoerova, C. Monnet and T. Temzelides, September 2009.
1092 Inflation and output volatility under asymmetric incomplete information by G. Carboni and M. Ellison,
September 2009.
1093 Determinants of government bond spreads in new EU countries by I. Alexopoulou, I. Bunda and A. Ferrando,
September 2009.
1094 Signals from housing and lending booms by I. Bunda and M. CaZorzi, September 2009.
1095 Memories of high inflation by M. Ehrmann and P. Tzamourani, September 2009.
46
ECB
Working Paper Series No 1112
November 2009
1096 The determinants of bank capital structure by R. Gropp and F. Heider, September 2009.
1097 Monetary and fiscal policy aspects of indirect tax changes in a monetary union by A. Lipiska
and L. von Thadden, October 2009.
1098 Gauging the effectiveness of quantitative forward guidance: evidence from three inflation targeters
by M. Andersson and B. Hofmann, October 2009.
1099 Public and private sector wages interactions in a general equilibrium model by G. Fernndez de Crdoba,
J.J. Prez and J. L. Torres, October 2009.
1100 Weak and strong cross section dependence and estimation of large panels by A. Chudik, M. Hashem Pesaran
and E. Tosetti, October 2009.
1101 Fiscal variables and bond spreads evidence from eastern European countries and Turkey by C. Nickel,
P. C. Rother and J. C. Rlke, October 2009.
1102 Wage-setting behaviour in France: additional evidence from an ad-hoc survey by J. Montorns and
J.-B. Sauner-Leroy, October 2009.
1103 Inter-industry wage differentials: how much does rent sharing matter? by P. Du Caju, F. Rycx and I. Tojerow,
October 2009.
1104 Pass-through of external shocks along the pricing chain: a panel estimation approach for the euro area
by B. Landau and F. Skudelny, November 2009.
1105 Downward nominal and real wage rigidity: survey evidence from European firms by J. Babeck, P. Du Caju,
T. Kosma, M. Lawless, J. Messina and T. Rm, November 2009.
1106 The margins of labour cost adjustment: survey evidence from European firms by J. Babeck, P. Du Caju,
T. Kosma, M. Lawless, J. Messina and T. Rm, November 2009.
1107 Interbank lending, credit risk premia and collateral by F. Heider and M. Hoerova, November 2009.
1108 The role of financial variables in predicting economic activity by R. Espinoza, F. Fornari and M. J. Lombardi,
November 2009.
1109 What triggers prolonged inflation regimes? A historical analysis. by I. Vansteenkiste, November 2009.
1110 Putting the New Keynesian DSGE model to the real-time forecasting test by M. Kolasa, M. Rubaszek
and P. Skrzypczyski, November 2009.
1111 A stable model for euro area money demand: revisiting the role of wealth by A. Beyer, November 2009.
1112 Risk spillover among hedge funds: the role of redemptions and fund failures by B. Klaus and B. Rzepkowski,
November 2009.
Worki ng PaPer S eri eS
no 1105 / november 2009
DoWnWarD nominal
anD real Wage
rigiDity
Survey eviDence
from euroPean
firmS
by Jan Babeck, Philip Du Caju,
Theodora Kosma, Martina Lawless,
Julin Messina and Tairi Rm
WAGE DYNAMICS
NETWORK